So they need a mechanism to decide what results are interesting. Maybe don't let anyone proof anything. Let the bot that sent a paper that got accepted to send

So they need a mechanism to decide what results are interesting. Maybe don't let anyone proof anything. Let the bot that sent a paper that got accepted to send two or three statements that they like to see proven. And at some point only allow proofs of things that have been requested in that way. There could also be a preference for short proofs.